Transformers Unidos: Eficacia De los Modelos Ensemble-ViT en Clasificación Automática de Flora Costarricense 🌿

This repository contains the code used to run the experiments for the paper "Transformers Unidos: Eficacia De los Modelos Ensemble-ViT en Clasificación Automática de Flora Costarricense 🌿". The paper explores the effectiveness of Vision Transformers ensemble models in comparison to an ensemble model with convolutional networks for the recognition of Costa Rican flora. The article can be found in the following link (in spanish).

Contents

• checkpoints: Directory to store model checkpoints.
• conv: Code for convolutional models.
• experiments: Scripts to run the experiments.
• metrics: csv with the data from the experiments as well as jupyter notebooks analysing the rusults.
• vit: Code for Vision Transformer models.
• configuration.py: Configuration constants for the experiments.
• data_modules.py and datasets.py: Code for data loading and preprocessing.
• helper_functions.py: Helper functions for the experiments.
• README.md: This file.

Usage

1. Clone the repository:

git clone https://github.com/Antonio-Tresol/vits_ensemble_cr_leaves.git

2. Install the required packages:

pip install -r requirements.txt

3. Get dataset from source
4. Update the configuration.py file with the dataset path and other desired settings.
5. Run the experiments using the scripts in the experiments directory.

Results

The results of the experiments are presented in the paper. The code in this repository can be used to reproduce the experiment.

Contact

If you have any questions, please contact Antonio Badilla-Olivas at anthonny.badilla@ucr.ac.cr.